Study area

Covering 9.2 million ha, the natural grasslands in the Kyrgyz Republic are the major feeding base for livestock such as cattle, sheep, horses, yaks, and goats. The grasslands are mostly located in the high mountains but also in steppes and semidesert areas (Kaufmann et al 2010). The field research was conducted in the Naryn oblast (district), which lies about 250 km southeast of the Kyrgyz capital, Bishkek, and has an area of 45,200 km² (Figure 1).

FIGURE 1 

Location of the study area. (Map by Thomas Breu, first published in Crewett 2012: 268, adapted for this case study)

Naryn was selected because of its variable terrain and the presence of all of the most common livestock species (Kaufmann et al 2010). The field research was carried out in different pasture management units in the Naryn rayon (an administrative subdivision of Naryn oblast). The pasture management units were categorized as follows, based on Ludi's (2003) description:

Within the Naryn rayon, pasture use problems (especially overgrazing) urgently need attention (Bussler 2010; Kaufmann et al 2010). According to Baibagushev (2011) there is an imbalance in the way livestock are placed in different pasture management units. Due to a lack of transport options and funds, livestock are often maintained close to settlements. This leads to high pressure on the winter and spring–autumn pastures. In contrast, the pressure on summer pastures is low.

For the field research, 3 pasture management units were chosen: 2 pastures that were primarily used as winter/spring–autumn pasture and 1 unit (located more than 2900 m above sea level) that was primarily used as summer pasture.

Methods

For the evaluation of pastures at the farm level with simple and limited equipment, the Visual Soil Assessment (VSA) (Shepherd 2009) was chosen. This method was developed in New Zealand with the aim of “providing a simple, inexpensive method to assess soil [and plant] quality semi-quantitatively, quickly and effectively” (Shepherd and Park 2003: 111). The VSA method is based on a weighted additive model, which includes indicators of soil properties (both static and dynamic) and plant performance. In this way it is designed to assess a wide range of different pasture conditions, which can be the result of multiple management regimes and land use pressures (Shepherd 2005). Since 2008, the method has been recommended by the FAO (2008). Our study was the first to use this method in a science-based approach in the Kyrgyz Republic (Figure 2).

FIGURE 2 

Application and verification of the VSA method in the field by Sabir Koshkonbaev (PhD student and colleague), Naryn Oblast, Kyrgyz Republic. (Photo by Peter Kirch, May 2013)

The VSA assessment uses 20 indicators, 10 for soil and 10 for vegetation. The score of each indicator is obtained by comparing a certain aspect of the site with a reference picture and criteria given in the VSA guide. Each picture and set of criteria corresponds to a certain condition of the indicator in question and is equivalent to a score between 0 and 2. A specific weighting factor between 1 and 3 is then applied through multiplication to each indicator score, leading to final indicator scores between 0 and 6. Indicator results are then summed separately for soil and plant conditions; the resulting scores indicate the overall soil and plant performance at each site. In cases of discrepancy between the 2, recommendations regarding possible changes in management practices can be derived. The VSA is standardized for all land use types, but it can also be adapted to account for specialized land uses such as pastoral grazing or annual crops (BioAgriNomics 2013).

In addition to the VSA method, 2 other field methods of visual site assessment with alternative indicator sets were tested during this study. These were the Muencheberg Soil Quality Rating (MSQR) (Mueller et al 2007) and the method described in the Monitoring Manual for Summer Pastures in the Greater Caucasus in Azerbaijan (referred to subsequently as the MMSP method) (Etzold and Neudert 2010). They were chosen on the basis of a literature search for methods similar to the VSA method in that they were indicator based, could be fully conducted in the field, and required only a small amount of equipment.

The MSQR assesses the current condition of the soil, including the medium-term soil hydrological, thermal, geological, and terrain conditions and the human impact (Mueller et al 2012), which are all considered soil-forming factors. For grassland, the ratings assume “a minimum level of accessibility and management” (Mueller et al 2007: 5). The result of the assessment is a semiquantitative measure, which can be interpreted as a rough estimate of the local crop yield potential.

The MMSP method is based on a preferential sampling design; pastures are assessed on the basis of representative sample areas. Information is gathered on the geographical situation, the soil, the vegetation, and the extent of erosion, and a visual appraisal is made of the state of the pasture. Figure 3 summarizes the indicators of all 3 methods and their scoring ranges.

FIGURE 3 

Visual site assessment methods applied during the study and their indicators.

In addition to the visual assessment methods, standard field measurements and laboratory-based analyses were conducted on the same pasture units to allow verification of the VSA indicator results. The leading selection criterion for these methods was simplicity of both the equipment and its application. This was essential due to the expected transportation challenges and working conditions in the mountainous terrain. Soil-related parameters were evaluated through the analysis of the soil samples, the measurement of the soil resistance to penetration, and the measurement of the water infiltration rate. In detail, the measured parameters were the following (the corresponding measurement standards or principles are given in brackets):

Based on the results, the overall soil condition was described.

Plant-related parameters were evaluated through the Klapp-Stählin and Braun-Blanquet methods (Dierschke 1994) and through a measurement of vegetative dry matter production. The choice of these methods also took the expected field conditions (mountainous terrain, limited transport infrastructure) and the limited field assessment time into account.

The 53 sample sites were chosen according to the catena concept (Eitel 2006). This was realized through the definition of catenas within the research units, taking the elevation gradient, moisture gradient, utilization gradient, and relief characteristics into consideration. Along the catenas, sample sites representative of certain topical vegetation units were selected.

All field assessment methods and field measurements were conducted twice during the research period (May to August 2013).

Data analysis and statistics

As a first step, the data obtained using the VSA method were interpreted with a method-specific evaluation system. During this process, each indicator was evaluated on its applicability in the field—with “applicability” defined as the extent to which the indicator could be determined in the field using the criteria in the VSA guide. Conditions in the research area such as relief, slope, and vegetation composition were particularly regarded. Each indicator was assigned to one of the following categories:

The “with difficulty” category included indicators for which there was inadequate assessment time during the field research period.

Second, the results of the first and second assessment periods were compared, with a special focus on comparison of results obtained in different pasture management units (winter/spring–autumn pasture and summer pasture).

Third, the 2 additional field methods of visual site assessment were also analyzed in the same way.

The fourth step was the comparison of the VSA results against the data obtained using standard field measurements and laboratory-based analyses. The VSA final site ratings as well as individual indicator scores were considered in this inquiry (see Table S5). To compare the results, the Kendall rank correlation coefficient τ (Kendall's tau) was calculated, as it is not sensitive for tied values (rank equalities). The software used during the process was the statistical package SPSS (IBM SPSS Statistics, version 22, 2013).

The results of the calculated correlation coefficient τ rank between −1 and +1. The following categories were used (based on Leyer and Wescher 2007) as no standard values for the interpretation of nonperfect values exist:

Applicability

The moderate applicability of the VSA method and the low overall correlation between the indicator ratings and the measured data sets showed that the VSA method can neither be fully applied nor verified as adapted to the prevailing site conditions in the research area. Besides the limited applicability of the different indicator assessment processes, the performance of the indicator rating in the research area was often assessed with difficulty. Some indicator ratings could not be assessed at all, including soil color and production costs to maintain stock-carrying capacity. In its current form and under the site conditions and pasture management system in the study area, the VSA assessment method seems therefore not to be suitable. This could be explained by the fact that the VSA method was developed for enclosed pasturelands in New Zealand on flat to rolling country, while the study area consists of free rangeland with a wide range of physiographic characteristics including mountainous terrain.

The alternative visual site assessment methods proved to be equally or more applicable under the given site conditions. The applicability of the MSQR method was estimated as equal to that of the VSA method. Its overall indicator composition seemed slightly better adapted to and applicable in the research area than the VSA method, but there was a need to perform additional device-based soil quality measurements to allow for a complete site rating. Thus this method would be fully applicable only by extension workers and experienced soil scientists.

The MMSP method was very well adapted to the prevailing site conditions and showed the best applicability of the 3 visual assessment methods (see Figure 4). This could be because this method was initially developed in an area with similar site conditions—primarily free rangeland in mountainous terrain in the Greater Caucasus.

Difficulties

During fieldwork and analysis, several difficulties with the visual assessment approach were identified. A major difficulty and a possible source of error was that all assessments and measurements were carried out only twice during the 4-month research period. This reduced the statistical significance of the results and the reliability of the data. This affected the evaluation quality, for example, the calculated correlations between the VSA indicator ratings and the measured data sets. To evaluate possible discrepancy between the measured data and the indicator outcomes with certainty, more extensive data sets are needed. In this study, even though the results of the measured soil and vegetation data were evaluated as plausible through a comparison with the results found in the literature, they do not necessarily reflect the actual site conditions with sufficient accuracy. This can be explained by the fact that the outcomes of many measurements of soil quality may vary because they depend on “the time of year the sample was taken for analysis, the nature of the season, the soil water content, the sampling depth, and the instrumentation and laboratory methodology used” (Shepherd 2003: 162). This reduces the reliability of the measured data.

A second difficulty was also connected to the approach of verifying the results of the visual assessments with the measured site condition values. The choice of standard field and laboratory-based methods for assessing soil and vegetation was based on “simplicity” of the field equipment required, its ease of use under the expected field conditions (limited transport infrastructure and mountainous terrain), and the limitation of research time needed (eg limited field assessment time). To what extent the resulting combination of chosen methods can precisely depict overall soil conditions and plant performance has to be questioned. Important aspects of the soil conditions—like aggregate size distribution, air permeability, macroporosity, and aggregate stability—were not directly assessable by these methods. The close relationships between the visual scores and the laboratory-based measurements of soil properties that Shepherd (2003) found were largely based on results for the above-listed aspects of soil condition. A comparison to these outcomes and the results obtained through the fieldwork could therefore only be partially performed.